EViews.com

I tried to SVAR identification using five variable (Government spending,Tax,GDP,CPI and Real interest rate) where 1990-2017 time series data. While I run VAR its prompted Maximum number of iteration exceed error. Then I increase maximum iterations and run again. In this time, it prompted Optimization may be unreliable error. can anyone help me to sort this thing out? I have attached work file and data set here.

Error codes:
1. Maximum number of iterations exceeded in "FREEZE (TABLE1) SGVAR1.SVAR(RTYPE=PATSR,NAMEA=PATA,NAMEB=PATB, F0=U)"
2. Optimization may be unreliable (first or second order conditions not met) in "FREEZE(TABLE1) SGVAR1.SVAR (RTYPE=PATSR,NAMEA=PATA,NAMEB=PATB, F0=U)".

Hello,

It's always good to try different types of initial values to see if one more effectively leads to convergence. The default normal distribution appears to work with your data (remove the "F0=U" option from the svar command).

Dear Matt.

Thank you for your reply. Could you please elaborate answer // It's always good to try different types of initial values to see if one more effectively leads to convergence// ?
I tried with removing F0=u but problem still same.

Ruwan

Hello,

Are you using EViews 10? If so, a different initial value choice seems to work better. Try an initial constant of one (use the "f0=1" option in the svar command).

Yes, I used eviews 10. It works when change f0=1.

Thank you